WHAT NEW IN X10.30 FRI

X 上的 AI 合集档案 · 22 个合集 · 61,645 条

deepseek-v4.1POST

用户反馈:V4.1 写代码好用,但日常 agent 幻觉明显

用户称 DeepSeek V4.1 编程表现不错,但作为通用日常 agent 幻觉率高,即使给了工具和严格规则仍坚持自身知识足够,和 Gemini 有同样毛病。

shaun
@Biggest11 天前收录原推发布于 2026.10.18
shaun @Biggest原推发布于 2026.10.18
在 X 上打开原推

I like DeepSeek V4.1 quite a lot for coding work, but man, as an actual daily general agent, its hallucination rate is so glaring. it's one of my complaints about Gemini too. they both like to insist their internal knowledge is enough even when given tools and strict rules of use

x.com/biggest/status/2111561368038207905采集于 2026.10.18 16:28